An improved unbiased method for diffspect quantification in epilepsy

Determining the region of seizure onset is of critical importance for treating medically intractable epilepsy. Comparisons between an ictal and interictal Single Photon Emission Computed Tomography (SPECT) images have been shown to be successful in localizing focal epilepsy. The Ictal-Interictal Subtraction Analysis by Statistical Parametric Mapping (ISAS) algorithm remains one the more successful algorithms for comparing these images. However ISAS is limited by its statistical design. This design introduces a scan order bias in the estimation of the normal variance of sequential SPECT images. We have corrected this bias by estimating the normal variance with a half-normal distribution. In this paper we present an updated algorithm (ISAS HN) based on the original ISAS algorithm with a corrected estimate of the normal variance and an open-source utility for ISAS HN.